100% chance of infection after 5 revisions.

[Discovernew]

So the n=2 on the relevant cohort? Two guys? That's it? It's impossible to generalize from a sample so small.

I agree about the 2/2, but if you would take everyone starting from the 3rd revision and above, it would be 22 guys and chance would still be more than 33%, which is huge.

The 3rd revision study with 33.3% rate is still powered only by 12 guys, so from a statistical standpoint the study is still highly underpowered.

For surgeons, this study can be useful in a medico-legal setting because it bolsters the argument that infections in such settings are driven by the number of prior revisions rather than surgical error!

It is not "only 12 guys" if you add up all guys with 3 revisions or more, its 22 guys, and all of them have 33% or higher. Which is huge. 33% is a 1/3rd or more chance to get infected.

Of course, i wish someone will come up with a new study that contradicts this one.

[lasthope2.0]

I wouldn't pool 3 different revision data from 3/4/5 into one single bucket as 22 guys. We cannot add the high risk of worser groups (4/5) to the lower risk of 3rd group unless the risk is uniform across all.

I think these percentage numbers (Ex: 2 patients and 100% risk) are like the monthly MAUDE statistics, they are only signal givers and hypothesis generators, but laden with lower quality statistical power from "lack of" or underpowered denominators.

As Dan said, by choosing a good surgeon, we bring down all risks to a lower number, as best as possible.

[Discovernew]

I wouldn't pool 3 different revision data from 3/4/5 into one single bucket as 22 guys. We cannot add the high risk of worser groups (4/5) to the lower risk of 3rd group unless the risk is uniform across all.

I think these percentage numbers (Ex: 2 patients and 100% risk) are like the monthly MAUDE statistics, they are only signal givers and hypothesis generators, but laden with lower quality statistical power from "lack of" or underpowered denominators.

As Dan said, by choosing a good surgeon, we bring down all risks to a lower number, as best as possible.

Thats why i said 33% OR more. Everyone with more than 3 revisions has more than 33% chance. So 33% is the baseline from there. There are 22 guys with 33% or more chance. The trend in the study seems clear, the higher amount of revisions, the higher chance of infection.

Of course i do hope an updated study will come out with more encouraging numbers than this one. This one seems to be from 2018.

[lasthope2.0]

I wouldn't pool 3 different revision data from 3/4/5 into one single bucket as 22 guys. We cannot add the high risk of worser groups (4/5) to the lower risk of 3rd group unless the risk is uniform across all.

I think these percentage numbers (Ex: 2 patients and 100% risk) are like the monthly MAUDE statistics, they are only signal givers and hypothesis generators, but laden with lower quality statistical power from "lack of" or underpowered denominators.

As Dan said, by choosing a good surgeon, we bring down all risks to a lower number, as best as possible.

Thats why i said 33% OR more. Everyone with more than 3 revisions has more than 33% chance. So 33% is the baseline from there. There are 22 guys with 33% or more chance. The trend in the study seems clear, the higher amount of revisions, the higher chance of infection.

Of course i do hope an updated study will come out with more encouraging numbers than this one. This one seems to be from 2018.

I'd love a newer study with better statistical power. About pooling, if a guy from 2nd group falls again in the 3rd and 4th group, you are adding him multiple times.

There's no problem with the trend. The percentage numbers are problematic due to poor statistical power.